An Auto-Indexing Method for Persian Text

This paper studies an approach to automatic indexing Persian context based on Persian grammatical rules in order to produce back-of-the-book index. Automatic indexing means automatically extract or select words from a document to create index. In this work, in order to present an approach for automatic indexing, SVM (Support Vector Machine) has been used to produce an intelligent system. The corpus has been applied is Bijankhan corpus which is a manually tagged Persian text collection. To evaluate proposed system, a book entitled Natural Low was considered as test set, while the index section of this book was done manually by human agent and compared with the automatic system. In this study, achieved precision and recall, were 53% and 90%, respectively.

Keywords:

Persian context , Automatic indexing , SVM